06:00 UTC Charta Aetatis Cyberpunk

Console Newsletter Pro Sign in Early access Register
Aa · Display
Themes
Type size
Density
MorningStar Security is in early access. The wiki is open to everyone; the news desks open with a code. Have a code, or want launch notice?
Vendors & Software

Language Models

Every vendor and product the wiki tracks. Vendor announcements are filed here instead of in the news.
Vendor · mature · 113 feeds Anthropic Anthropic is an artificial intelligence research company founded in 2021 and headquartered in the United States. HQ CountryUnited StatesFounded2021 Product · stub · 1 feed big-pickle Free stealth LLM model on OpenCode Zen achieving 50.8% task resolve rate on SWE Atlas Codebase QnA benchmark Categorylanguage-model Product · stub · 1 feed Cyber-OSINT Locally-runnable MoE language model fine-tuned for OSINT, CTI, threat attribution and IoC analysis VendorDeepHatCategorylanguage model Product · stub · 1 feed Ember-1 Specialized LLM from Fireworks Research derived from Kimi K3, achieving same quality with 40% fewer tokens through efficient reasoning training VendorFireworks ResearchCategorylanguage model Product · stub · 1 feed Epilude Model 4 Fourth-generation cleanup model for Local Mode, built on Qwen, 56% reduction in critical errors vs Model 3 Categorylanguage model Product · growing · 17 feeds Gemini 4 Argon Frontier AI model for software engineering, enterprise knowledge work, and cybersecurity defense with 1M token output limit VendorGoogleCategorylanguage model Product · growing · 10 feeds Gemma 4 Google's 26-billion-parameter open-weights mixture-of-experts language model VendorGoogleCategorylanguage model Product · stub · 1 feed Gemma-4-26B-A4B-StyleTune-V2 26B parameter language model with modified output projection layer to reduce clichéd language in creative writing while preserving reasoning VendorGrypheCategorylanguage model Vendor · stub · 1 feed gnani Developer of the Gnani Evon language model series. Product · growing · 3 feeds GPT-2 VendorOpenAICategorylanguage-model Product · growing · 5 feeds GPT-5.4 mini Categorylanguage model Product · growing · 6 feeds GPT-6.1 Sol GPT-6.1 Sol is reported to be a language model offered by OpenAI as a cloud-based API. VendorOpenAICategorylanguage model Product · stub · 1 feed Inception Mercury 2.5 VendorInceptionCategorylanguage model Vendor · growing · 11 feeds Kimi Product · stub · 1 feed Kolibri English-German Mixture-of-Experts Transformer with 78B total parameters, 3B active, 1M token context, optimized for sovereign mission-critical work in… VendorAleph AlphaCategorylanguage-model Product · stub · 1 feed Laguna S 2.1 118B MoE model with 8B activated parameters, 1M token context, optimized for long-horizon coding tasks VendorPoolsideCategorylanguage model Product · growing · 87 feedssheet Large Language Models They are described as developed by vendors including OpenAI and, among others, Anthropic, Google, Meta and xAI. VendorOpenAI, MicrosoftCategorylanguage model Product · stub · 1 feed LatamGPT Open-source large language model developed for and by Latin America, built on Meta's Llama foundation, trained on AWS infrastructure CategoryLarge Language Model (LLM) Product · stub · 1 feed LFM2.5-1.2B-Instruct-Heretic 1.2 billion parameter language model for local AI inference on mobile devices Categorylanguage model Product · stub · 1 feed LFM2.5-2.6B 2.6B parameter hybrid model optimized for on-device deployment with 128K context window and agentic capabilities VendorLiquid AICategorylanguage-model Product · stub · 1 feed LittleLearner Language models trained exclusively on K–5 curriculum to study knowledge acquisition vs. elicitation Categorylanguage model Product · stub · 1 feed Mercury 2 VendorInception LabsCategorylanguage model Product · stub · 1 feed Mercury Voice Categorylanguage model Product · stub · 2 feeds mini-AGI Continual learning byte-level language model trainable from scratch on 8GB VRAM GPU with dynamic architecture assembly and on-disk weight paging. Categorylanguage model Product · stub · 1 feed Needle 2 45M-parameter open agentic LLM compressed to 14MB for on-device tool calling and structured extraction on resource-constrained devices VendorPebbleCategorylanguage model Product · stub · 1 feed Neutrino-1 8B 8.19B-parameter decoder-only transformer with proprietary ternary-family weight compression, shipping as 3.88 GB file, runs on datacenter GPUs, laptop… VendorAlibaba CloudCategorylanguage model Vendor · mature · 116 feeds OpenAI OpenAI is an artificial intelligence research and deployment organization. Founded in 2015, it is headquartered in the United States. HQ CountryUnited StatesFounded2015 Product · stub · 1 feed Open-Source LLM Categorylanguage model Product · stub · 1 feed PSSA Non-transformer language model written in Rust with recurrent state-space core, episodic memory bank, and plastic weights VendorSparticle62opsCategorylanguage model Tool · stub · 2 feeds Qwen2.5-3B-Instruct Categorylanguage model Product · growing · 2 feeds Qwen2.5-7B-Instruct Categorylanguage model Product · stub · 4 feeds Qwen3.6-27B Categorylanguage model Product · stub · 1 feed Qwen3.6-27B-Fable-Fusion 27B open-source LLM achieving 700+ ARC-C scores with 256k context, vision, and decensoring VendorDavidAUCategorylanguage model Product · stub · 3 feeds Qwen 3.8 Flash Categorylanguage model Product · mature · 7 feeds Qwen 3.8-Flash-Next Large language model with 125B parameters and 6B active parameters, announced for imminent release VendorAlibabaCategorylanguage model Product · stub · 1 feed Qwen 3.8 Max Preview Categoryadvanced language model Product · stub · 1 feed Sophontic AI model achieving reasoning performance exceeding models 60× its size through geometric density optimization VendorSophonticCategorylanguage model Product · stub · 1 feed StartLux-V1.0-27B-Preview 27 billion parameter local AI model achieving competitive performance on agent tasks Categorylanguage model Product · stub · 1 feed TwIL-8B 8.79B-parameter text-generation model post-trained from IBM Granite 4.2 for contest math, code, and instruction following VendorwebAI-OfficialCategorylanguage model

Category wiki page →

That's the Front Page for today

Tomorrow's desk, in your inbox by 06:00 UTC. The week's best stories, every Sunday. The month that mattered, on the 1st.

Drawn from the 519 sources the paper read this month. Text and HTML, one-click unsubscribe, free.

How often
Your desks